OpenAI's Latest Model Aims to Master Computer Tasks, Sparking AGI Debate
OpenAI has introduced GPT-6 Astra, a new AI model it says outperforms humans in navigating computers, writing code, and solving complex math. The company plans to roll it out to paying customers soon, starting with enterprise users. Executives, including president Greg Brockman, believe this release could mark the beginning of the artificial general intelligence era.
The rollout strategy prioritizes paying enterprise clients through the Daybreak early access program, with consumer tiers like Plus, Pro, and Business following shortly after. OpenAI has not confirmed whether free ChatGPT users will gain access. This commercial push comes as the company prepares for a planned initial public offering, while rival Anthropic also works toward its own IPO.
Safety claims center on monitoring the model's internal chain-of-thought reasoning. Chief scientist Jakub Pachocki acknowledged that as capabilities grow, ensuring alignment becomes harder, and the company may withhold scaling if monitoring confidence drops. This highlights a tension between rapid advancement and maintaining safety assurances.
GPT-6 Astra could significantly reshape digital labor, potentially automating routine computer tasks like scheduling, form-filling, and coding, which may affect administrative and technical workers. For consumers, it could simplify online interactions but also raise privacy and reliability concerns. The AGI framing may intensify public debate